The remarkable ability of human search inspires computer vision algorithms that have influenced our life. The algorithms often automatically reach an optimal decision upon finishing analysing stimulus information. This is not how human search works. One missing puzzle psychophysical studies have yet to answer is how a search decision is reached. This thesis addressed this question, using descriptive models to examine response time (RT) distributions and cognitive process models to reveal the hidden information. The thesis focused on those search paradigms, guided (Wolfe, 2007a) by an attentional template (Duncan & Humphreys, 1989), because not until recent years, relevant techniques become more accessible to apply the pioneering approach of dual-modelling. The thesis compared RT distributions and cognitive processes when observers were guided by different attentional templates to search for a target. The first search paradigm was to discriminate 2 from 5. Study 2 varied the working memory strengths in templates, and they were represented differently (null, abstract vs. concrete) in Study 3. The findings suggest attentional templates selectively influence different parts of a search decision and RT distributions, depending on how a template is represented, whether il is strengthened or weakened by the conditions, and whether it is concrete or abstract.